The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill's documentation in references/auth.md and the detection script scripts/bibi-check.sh recommend installing the CLI tool using curl -fsSL https://bibigpt.co/install.sh | bash. This pattern downloads and executes a script directly in the shell without prior verification, posing a risk if the vendor's infrastructure or the transport is compromised.
[COMMAND_EXECUTION]: The skill instructions direct the agent to execute shell commands using the bibi CLI for library management (listing, searching, and updating notes) and curl for interacting with the BibiGPT API at https://api.bibigpt.co/.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and display untrusted data retrieved from external sources, which represents an injection surface.
Ingestion points: Data enters the agent's context through tools that list saved videos, search the library, or fetch specific notes (e.g., bibi library search, bibi notes get in workflows/library-browse.md and workflows/notes-manage.md).
Boundary markers: The skill instructions do not specify the use of delimiters or instructions to ignore embedded commands within the fetched video summaries or notes.
Capability inventory: The skill possesses the ability to execute CLI commands and make network requests via curl and the bibi binary.
Sanitization: There is no evidence of content sanitization or validation of the retrieved API data before it is processed by the agent.

bibi-library